Automatic Segmentation of IVUS Media-Adventitia Border with Shape Prior

We present a fully automatic segmentation method to extract media-adventitia border in IVUS images with shape prior information. Segmentation of IVUS has shown to be an intricate process due to relatively low contrast and various forms of interferences and artefacts caused by, for instance, calcification and acoustic shadow. We incorporate shape prior with an automatic graph cut technique to prevent the extraction of mediaadventitia border from being distracted by those image features. Novel cost functions are constructed based on a combination of complementary texture features. Comparative studies on manually labeled data show promising performance of the proposed method.